INDEX

Explanations

This new invention

The neuron fires on code fragments defining or using regular expressions (e.g. the “/^([^ \r.]*)…” pattern), so it’s looking for regex syntax in code.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 hunne

-1.28

iffany

-1.23

kimono

-1.23

 图解

-1.23

 eenen

-1.19

jsh

-1.19

 dezelve

-1.17

appli

-1.16

 gekozen

-1.13

 mauva

-1.12

POSITIVE LOGITS

1.70

..

1.18

1.17

new

1.16

new

1.12

–

1.09

or

1.09

 while

1.06

..

1.05

Activations Density 0.000%